Hospitalizations for suspected dengue in Puerto Rico, 1991-1995: Estimation by capture-recapture methods

Capture-recapture estimations compare the results of 2 or more independent surveillance systems for the same event, and by measuring the degree of ove rlap between them, provide an estimate of the total number of events, and t herefore the completeness of ascertainment in each system. The Puerto Rico Department of Health and the Dengue Branch of the Centers for Disease Contr ol and Prevention (CDC) monitor dengue activity in Puerto Rico through 2 di stinct surveillance systems: diagnostic specimens From patients with suspec ted dengue and infection control nurses' reports on patients hospitalized f or suspected dengue. The patient listings from these systems were used in a 2-sample, capture-recapture calculation to estimate the total number of pe rsons with suspected dengue hospitalized from 1991 to 1995. The laboratory positivity rate for suspected dengue cases who submitted appropriately time d serum samples in those years ranged from 72.1% to 81.2%. The laboratory-b ased (diagnostic sample) surveillance system (routinely used to monitor hos pitalizations for suspected dengue) detected an average of 1,197 hospitaliz ed cases during non-epidemic years, and 4,329 cases during;the epidemic yea r of 1994. The detection rate of this system averaged 42% of the numbers de rived by the capture-recapture method. In non-epidemic years, an estimated average of 2,791 patients (range = 1,553-3,481) was estimated to have been hospitalized with a clinical diagnosis of dengue, compared with 9,479 durin g 1994. These results demonstrate the under-detection inherent in passive s urveillance systems for hospitalized cases of suspected dengue, and illustr ate the value of capture-recapture techniques to better estimate the true i ncidence of hospitalizations for this disease.